On the role of single regiodefects and polydispersity in regioregular poly(3-hexylthiophene): Defect distribution, synthesis of defect-free chains, and a simple model for the determination of crystallinity

Article abstract of DOI:10.1021/ja210871j

Identifying structure formation in semicrystalline conjugated polymers is the fundamental basis to understand electronic processes in these materials. Although correlations between physical properties, structure formation, and device parameters of regioregular, semicrystalline poly(3-hexylthiophene) (P3HT) have been established, it has remained difficult to disentangle the influence of regioregularity, polydispersity, and molecular weight. Here we show that the most commonly used synthetic protocol for the synthesis of P3HT, the living Kumada catalyst transfer polycondensation (KCTP) with Ni(dppp)Cl₂ as the catalyst, leads to regioregular chains with one single tail-to-tail (TT) defect distributed over the whole chain, in contrast to the hitherto assumed exclusive location at the chain end. NMR end-group analysis and simulations are used to quantify this effect. A series of entirely defect-free P3HT materials with different molecular weights is synthesized via new, soluble nickel initiators. Data on structure formation in defect-free P3HT, as elucidated by various calorimetric and scattering experiments, allow the development of a simple model for estimating the degree of crystallinity. We find very good agreement for predicted and experimentally determined degrees of crystallinities as high as ～70%. For Ni(dppp)Cl₂-initiated chains comprising one distributed TT unit, the comparison of simulated crystallinities with calorimetric and optical measurements strongly suggests incorporation of the TT unit into the crystal lattice, which is accompanied by an increase in backbone torsion. Polydispersity is identified as a major parameter determining crystallinity within the molecular weight range investigated. We believe that the presented approach and results not only contribute to understanding structure formation in P3HT but are generally applicable to other semicrystalline conjugated polymers as well.

Full text of DOI:10.1021/ja210871j

Article
pubs.acs.org/JACS
On the Role of Single Regiodefects and Polydispersity in
Regioregular Poly(3-hexylthiophene): Defect Distribution, Synthesis
of Defect-Free Chains, and a Simple Model for the Determination of
Crystallinity
Peter Kohn,† Sven Huettner,‡ Hartmut Komber,§ Volodymyr Senkovskyy,§ Roman Tkachov,§
Anton Kiriy,§ Richard H. Friend,‡ Ullrich Steiner,† Wilhelm T. S. Huck,⊥,▽ Jens-Uwe Sommer,§
and Michael Sommer*,⊥,#
†Biological and Soft Systems, Department of Physics, University of Cambridge, J J Thomson Avenue, Cambridge CB3 0HE, U.K.
‡Optoelectronics Group, Cavendish Laboratory, University of Cambridge, J J Thomson Avenue, Cambridge CB3 0HE, U.K.
§Leibniz- Institute of Polymer Research Dresden e.V., Hohe Strasse 6, 01069 Dresden, Germany
⊥Melville Laboratory for Polymer Synthesis, Lensfield Road, Cambridge CB2 1EW, U.K.
▽Radboud University Nijmegen, Heyendaalseweg 135, 6525 AJ Nijmegen, Netherlands
S
* Supporting Information
ABSTRACT: Identifying structure formation in semicrystal-
line conjugated polymers is the fundamental basis to
understand electronic processes in these materials. Although
correlations between physical properties, structure formation,
and device parameters of regioregular, semicrystalline poly(3-
hexylthiophene) (P3HT) have been established, it has
remained difficult to disentangle the influence of regioregu-
larity, polydispersity, and molecular weight. Here we show that
the most commonly used synthetic protocol for the synthesis
of P3HT, the living Kumada catalyst transfer polycondensation
(KCTP) with Ni(dppp)Cl2 as the catalyst, leads to regioregular chains with one single tail-to-tail (TT) defect distributed over the
whole chain, in contrast to the hitherto assumed exclusive location at the chain end. NMR end-group analysis and simulations are
used to quantify this effect. A series of entirely defect-free P3HT materials with different molecular weights is synthesized via
new, soluble nickel initiators. Data on structure formation in defect-free P3HT, as elucidated by various calorimetric and
scattering experiments, allow the development of a simple model for estimating the degree of crystallinity. We find very good
agreement for predicted and experimentally determined degrees of crystallinities as high as ∼70%. For Ni(dppp)Cl2-initiated
chains comprising one distributed TT unit, the comparison of simulated crystallinities with calorimetric and optical
measurements strongly suggests incorporation of the TT unit into the crystal lattice, which is accompanied by an increase in
backbone torsion. Polydispersity is identified as a major parameter determining crystallinity within the molecular weight range
investigated. We believe that the presented approach and results not only contribute to understanding structure formation in
P3HT but are generally applicable to other semicrystalline conjugated polymers as well.
conditions, e.g., the solvent.15 Crystallization from poor
1. INTRODUCTION
solvents can lead to very long whiskers, in which the chain
can fold back and forth several times along the whisker axis.16
Besides processing, molecular parameters such as molecular
weight,17−23 nature of chain ends,24,25 polydispersity,26 and
regioregularity (RR), i.e., the content of head-to-tail (HT)
couplings,27−32 influence the optoelectronic properties. The
effects of microstructure and ordering on the device perform-
ance of P3HT films have been addressed by numerous
studies.33−37 Generally, a high RR is important to improve
Regioregular poly(3-hexylthiophene) (rrP3HT) has evolved as
a ubiquitous material for organic electronic devices due to its
easy availability and tendency to form semicrystalline lamellar
microstructures.1−7 The power conversion efficiencies of
organic photovoltaic devices (OPV) incorporating blends of
rrP3HT and fullerene derivatives have been gradually increased
to high values up to 6.48% during the last years.8−14 Films of
pristine rr-P3HT are also used in organic field effect transistors
(OFETs), which typically yield appreciably high charge carrier
mobilities of 10−2 to 10−1 cm2/(V s). The performance of such
thin film devices critically depends on the microstructure and
crystallinity, which in turn is governed by the processing
Received: November 18, 2011
Published: February 14, 2012
© 2012 American Chemical Society
4790
dx.doi.org/10.1021/ja210871j | J. Am. Chem. Soc. 2012, 134, 4790−4805

Journal of the American Chemical Society
Article
crystallinity and thus device performance: It has been shown
that crystal thickness38 and crystallinity increase with increasing
RR27,28,33,39 and that a large content of tail-to-tail (TT) and
head-to-head (HH) couplings, as prevalent in regiorandom-
P3HT (rran-P3HT), can lead to the complete absence of main-
chain crystallinity.31,40 Even relatively small variations of RR
between 90% and 96% can cause major changes in device
performance.28,33 These results were explained by the fact that
higher RRs can improve π−π stacking and thus two-
dimensional charge transport.28,33 More recent results suggest
that photovoltaic blends composed of P3HT may not
necessarily require a high RR for optimum device perform-
ance.29,31 Despite these general trends, it has remained difficult
to assign changes in microstructure or crystallinity to one
particular molecular parameter. Therefore, it is still unclear
which particular aspect of the microstructure relates to a certain
macroscopic electronic property. The difficulty to establish
such relationships may arise from imperfect synthetic control
over starting materials but also from the inability to isolate the
influence of one single molecular parameter on macroscopic
properties within a given set of polymers.
One of the most widely used synthetic route to synthesize
rrP3HT is the Kumada catalyst transfer polycondensation
(KCTP) developed by McCullough et al.5 and Yokozawa et
al.41 This method utilizes the commercially available catalyst
(1,3-bis(diphenylphosphino)propane)dichloronickel (Ni-
(dppp)Cl2) 1 and leads to highly regioregular P3HT
(rrP3HT) with low polydisperities between 1.1 and 1.3.
Initiation proceeds from an in situ-formed tail-to-tail (TT)
dimer formed from Ni(dppp)Cl2 and two monomer molecules,
from which propagation of rrP3HT occurs.42,43 This TT
coupling defect, arising from the structure of the catalyst, is
therefore inevitably incorporated into the otherwise regiore-
gular chain. While it was until now assumed that the TT
initiating dimer is exclusively located at the beginning of the
P3HT chain, a recent study suggests that this might only
partially be true: Tkachov et al. observed “random ring walking”
when polymerizing P3HT from the external model initiator 4-
bromobenzene−Ni(dppe)-Br and found P3HT with an internal
phenyl group as the major product.44 If the same mechanism
was operative during the polymerization via commercially
available Ni(dppp)Cl2, the TT defect would be located
anywhere within the linear P3HT chain. Thus, only the
distribution of the TT defect within the chain fully characterizes
a P3HT macromolecule.
ends of the P3HT chain occurs. The resulting distribution
of the TT defect within the backbone is characterized by
a combination of NMR end group analysis and computer
simulations. We further present the synthesis of a series of
entirely defect-free P3HT materials with 100% RR via new
soluble nickel initiators, in which the TT defect is eliminated.
The crystallization behavior of the two series, i.e., well-defined,
conventionally polymerized samples and well-defined, defect-
free samples, is comparatively studied by X-ray scattering and
calorimetric and optical experiments. We finally present a
simple model for the determination of crystallinity based on the
molecular weight distribution. This model quantitatively
accounts for experimentally observed crystallinities in defect-
free P3HT and corroborates the strong influence of
polydispersity on crystallinity. The combination of simulated
TT defect distributions, experimental crystallinities, and the
model suggests that partial inclusion of tail-to-tail units into the
crystal lattice is tolerated.
2. EXPERIMENTAL SECTION
Synthetic Procedures. General Procedures and Materials.
Solvents and starting materials were obtained from Aldrich
unless otherwise noted and used as received. Commercially
available dry THF was distilled over sodium and benzophenone
ketyl and stored under argon. Diethylbipyridylnickel (Et2Ni-
(bipy)),49 2-bromo-5-chloromagnesio-3-hexylthiophene,50 and
the conventionally polymerized samples 2a−e (route A)42 were
synthesized through standard literature procedures.
Initiators. Initiator Ni(dppe)ThBr (3a): In a glovebox, 217 mg
(0.88 mmol) of 2-bromo-3-hexylthiophene and 8 mL of dry THF were
placed into a 100 mL flask and sealed with a septum. The flask was
transferred outside, and the solution was cooled to −15 °C. A 200 mg
amount of Et2Ni(bipy) (200 mg, 0.73 mmol) was added dropwise
in 10 mL of THF via syringe. The initial green color changed to
yellow and finally red, indicating completion of reaction. A 350 mg
amount of 1,2-bis(diphenylphosphino)ethane (dppe) was added in
10 mL in one portion, upon which a slight color change from red
to orange-red was observed. A small portion of THF was evaporated,
and the mixture was transferred back to the glovebox. Filtration into
dry diethyl ether/n-hexanes 1:2 yielded a yellow precipitate, which
was further crystallized at −20 °C overnight. The raw product was
redissolved, filtered, and reprecipitated to yield 326 mg (0.46 mmol,
63%) of pure, shiny yellow crystals. The 1H and 31P NMR spectra are
depicted in Figure SI-1, Supporting Information. For atom numbering,
see Scheme 2. 1H NMR (THF-d8, 303 K): 8.33 (m, 2H, HAr), 8.27 (m,
2H, HAr), 7.68 (m, 2H, HAr), 7.6−7.45 (6H, HAr), 7.37 (3H, HAr), 7.30
2
(t, 1H, HAr), 7.08 (2H, HAr), 7.07 (dd, JHH = 4.8 Hz, JPH = 2.1 Hz,
1H, Ha), 6.84 (m, 2H, HAr), 6.40 (dd, 2JHH = 4.8 Hz, JPH = 1.4 Hz, 1H,
Hb) 2.65−2.45 (2H, Hc, HαCH2), 2.4−2.2 (2H, Hc), 1.94 (m, 1H,
Independent of the exact location of this TT defect, its
influence on microstructure is not clear and cannot be
separated from other molecular parameters. It has been
suggested that the apparent molecular weight dependence of
crystallinity in P3HT45,46 might be due to terminal TT defects
which were considered as noncrystallizable groups.46 Polydis-
persity can affect crystallinity as well,46 similar to n-alkanes:47,48
From a crystal layer comprising shorter and longer chains in a
random manner, amorphous ciliae of the longer chains
protrude, which limits crystallinity. The situation may become
even more complicated if the TT defect is not exclusively
located at the chain ends, as the additional question arises if
nonterminal TT defects can be incorporated into the crystal
lattice. Therefore, it remains challenging to separate the effects
of regioregularity and polydispersity unless truly monodisperse
or entirely defect-free samples are studied.
H
H
αCH2), 1.61 (m, 1H, Hc), 1.54 (m, 1H, HβCH2), 1.35−1.15 (7H,
βCH2, (CH2)3), 0.91 ppm (t, 3H, CH3). 31P NMR (THF-d8, 303 K):
58.04 (d, JPP = 32.9 Hz, P2), 42.58 ppm (d, JPP = 32.9 Hz, P1).
Initiator Ni(dppp)ThBr (3b): Same procedure as for 3a yielded 301
mg (0.42 mmol, 57%) of shiny golden crystals. The 1H and 31P NMR
spectra are depicted in Figure SI-2, Supporting Information. For atom
1
numbering, see Scheme 2. H NMR (THF-d8, 303 K): 8.25 (br, 2H,
HAr), 8.09 (br, 2H, HAr), 7.81 (br, 2H, HAr), 7.44 (br, 4H, HAr), 7.35
(br, 5H, HAr), 7.17 (br, 1H, HAr), 7.00 (br, 4H, HAr), 6.85 (d, JHH = 4.8
Hz, 1H, Ha), 6.14 (d, JHH = 4.8 Hz, 1H, Hb), 2.85 (br, 1H, HαCH2),
2.64 (br, 1H, Hc), 2.40 (br, 1H, Hc), 2.26 (br, 1H, Hc), 2.04 (br, 3H,
Hc, Hd, HαCH2), 1.69 (br, 1H, HβCH2), 1.44 (m, 1H, Hd), 1.35−1.25
(7H, HβCH2, (CH2)3), 0.91 ppm (t, 3H, CH3). 31P NMR (THF-d8, 303
K): 18.45 (d, J = 62.5 Hz, P2), −3.67 ppm (d, J = 62.5 Hz, P1).
General Polymerization Procedure for Syntheses of Defect-Free
P3HT. Syntheses of defect-free P3HT 4a−l (route B) was carried out
under argon in single flasks equipped with a stir bar and a septum.
The concentration of 2-bromo-5-chloromagnesio-3-hexylthiophene
in THF was 0.067 mmol/mL. Different amounts of monomer
We show here that in Ni(dppp)Cl2-initiated polymerizations
of P3HT bidirectional growth and reinsertion of Ni(0) at both
4791
dx.doi.org/10.1021/ja210871j | J. Am. Chem. Soc. 2012, 134, 4790−4805

Journal of the American Chemical Society
Article
active monomer.42,43 Until now it was assumed that chain
propagation proceeds unidirectionally from one side of this
initiating TT-dimer.5,43 We show that, in accordance with
recent results from Tkachov et al.,44 insertion of Ni(0) occurs
at both sides of the initially formed symmetric tail-to-tail dimer
at all times of polymerization, resulting in bidirectional growth
of the P3HT chain. This leads to an effective “migration” of the
TT defect into the backbone from the very beginning and
during all times of polymerization, and finally leads to a
distribution of the TT defect within the chain (Scheme 1a).
(for [M]/[I] ratios, see Table SI-1, Supporting Information) were
added to solutions of 3a or 3b (∼20−40 mg/mL). Polymerizations
were carried out at room temperature for several minutes (3a) and for
30−45 min (3b). The polymerization was quenched with 5 M HCl;
care was taken to avoid precipitation. After the mixture was stirred
further for several minutes, methanol was added to precipitate the
polymer and the mixture was filtered into an extraction thimble. The
solids were extracted with methanol and chloroform, respectively, and
the chloroform solution was concentrated to dryness under reduced
pressure. Methanol was added, and the solids were collected and dried
in vacuo. A compilation of the physical properties of all polymers is
given in the Supporting Information. The properties of four polymers
studied further in detail are summarized in Table 1.
1
End group analysis of the H NMR spectra was used to
distinguish between chains with a terminal defect and chains
with an internal one. Figure 1a shows the spectrum of the low
molecular weight sample 2a with all important assignments. As
evidenced by correlations to 2-bromothiophene carbons at
about 108 ppm in the HMBC spectrum, all signals in the 6.86−
6.81 ppm region arise from protons of bromine end groups
(Figure SI-3, Supporting Information). The assignment of the
singlet at 6.86 ppm (H2) and the singlet of equal intensity at
6.93 ppm (H3) to the TT end-group carrying bromine was
accomplished by a combination of 1D and 2D NMR
techniques. The 13C signal assignment is given in the
Supporting Information. The well-known signal of bromine
caps of regioregular end groups (H10) appears at 6.84 ppm. The
signals at the low frequency side of H10 also result from
bromine end-groups and are therefore thought to arise from
bromine chain ends with the TT defect being at the second and
third position. The signals at 6.90 and 6.80 ppm show a
broadening due to coupling to the α-CH2 group of the hexyl
residue typical for H termination. Whereas the first signal is
known as H7 of regioregular P3HT, the low field signal is
assigned to H4 of a terminal TT group carring hydrogen at the
end. This signal indicates that also the Br-TT end can be
converted into an initiating side by random ring walking of
Ni(0), which starts chain growth at the other side of the
polymer chain. The thus-produced TT defect within the chain
is characterized by a proton signal at 7.00 ppm, in accordance
with the well-known chemical shift for a HT-TT triad.54,55 This
peak on top of the backbone signal is hard to quantify (see
arrow in Figure 1a). However, by comparing the signal intensity
of H2 with the intensity of the other bromine end group signals
between 6.85 and 6.81 ppm, we find that the percentage of
chains having the TT defect at the end (y = 0, see Scheme 1,
denoted as TT-P3HT) is already low for small molecular
weights of several kg/mol and further decreases with increasing
degree of polymerization. Figure 2b displays this behavior. Only
28% of the chains in the low molecular weight control sample
2a (Mn,GPC = 4.2 kg/mol, DPNMR = 17) carry the TT unit at the
chain end. This fraction decreases to 18% for sample 2e
(Mn,GPC = 21.2 kg/mol, DP = 76). The decrease of the chains
having the terminal defect with increasing molecular weight is
more pronounced compared to the analogous bifunctional
phenyl-based initiator from Tkachov et al., which can be
understood as a result from the higher reactivity of the
bromine−thiophene bond toward Ni(0) in comparison to the
bromine−phenyl bond.44 The same end group signals were also
used to estimate the molecular weight, which is more precise
and consistent than GPC or MALDI-based values.34 Values for
the degree of polymerization therefore refer to NMR-based
values, unless otherwise stated.
Table 1. Physical Properties of the Four Samples Studied in
a
Detail
1TT-
0TT-
0TT-
1TT-
P3HT-43 P3HT-45 P3HT-65 P3HT-76
reference to Supporting
Information
2c
4i
4f
2e
TT defect
1
0
0
1
Mn (GPC) [kg/mol]
PDI (GPC)
11.8
1.13
6.9
13.4
1.11
7.5
17.2
1.26
10.5
1.028
63
21.2
1.28
b
Mn,MALDI [kg/mol]
PDI (MALDI)
DP (MALDI)
−
−
b
1.022
41
1.020
45
b
−
DP (NMR)
43
45
65
76
d100 [nm] (190 °C)
d002/020 [nm] (190 °C)
1.82
0.38
1.81
0.38
1.82
0.38
1.83
0.38
a
1TT-P3HT-43 denotes a sample with one distributed TT defect
(initiated by Ni(dppp)Cl2) with a degree of polymerization (DP) of
43. Mn,GPC was obtained from calibration with polystyrene standards,
Mn,MALDI was calculated from the peak intensities, DPMALDI
=
b
Mn,MALDI/(166.3 g/mol). The high molecular weight of 1TT-P3HT-
76 resulted in a noisy MALDI spectrum and therefore was not used to
calculate molecular weights.
3. RESULTS AND DISCUSSION
3.1. Defect Distribution of Ni(dppp)Cl2-Initiated Poly-
(3-hexylthiophene). Ni(dppp)Cl2 1 or Ni(dppe)Cl2 are
common catalysts used for the synthesis of regioregular
P3HT via KCTP. This method developed by McCullough
et al. uses a 70:30 to 80:20 mixture of “regular” 2-bromo-5-
chloromagnesio-3-hexylthiophene and “reversed” 5-bromo-2-
chloromagnesio-3-hexylthiophene, respectively,42 while Yokozawa
and co-workers use pure “regular” 2-bromo-5-chloromagnesio-
3-hexylthiophene as the active monomer.43,50 Catalysts such
as 1 are usually used, which are highly selective toward
polymerization of the regular monomer, and thus the two
methods result in the same chain topology and regioregularity.
The “reversed monomer” 5-bromo-2-chloromagnesio-3-
hexylthiophene only polymerizes under specific conditions
such as additional lithium chloride,51−53 which causes a lower
RR. However, such conditions were not used here, and
therefore the control materials 2a−e obtained from 1 are fully
regioregular P3HT chains with exactly one tail-to-tail (TT)
defect. With the regioregularity (RR) being defined as the
percentage of head-to-tail linkages, we can expect RR values
from these protocols according to RR = (DP-2)/(DP-1), where
DP is the degree of polymerization. The single TT regiodefect
arises from the initial reduction of Ni(dppp)Cl2 with 2 equiv of
1H NMR end group analysis can only distinguish between
the two cases where the TT unit is either located at the end of
the chain or anywhere else. In order to reveal the distribution of
4792
dx.doi.org/10.1021/ja210871j | J. Am. Chem. Soc. 2012, 134, 4790−4805

Journal of the American Chemical Society
Article
Scheme 1. (a) Common Procedure for Kumada Catalyst Transfer Polycondensation Using Ni(dppp)Cl2, Resulting in
a
Regioregular P3HT 2a−e with One TT Defect Being Distributed within the Chain. (b) Hexylthiophene-Based Nickel Initiators
b
NiL2ThBr 3a,b (L2= dppe, dppp) for the Synthesis of Defect-Free P3HT
a
The fraction of chains with a terminal TT unit is a function of the degree of polymerization.
b
Samples 4a−f and 4g−i are polymerized starting from 3a and 3b, respectively.
1
1
Figure 1. H NMR end group analysis of P3HT. (a) Proton assignments of Ni(dppp)Cl2-initiated P3HT and H NMR spectrum (regions) of
sample 2a (DP = 18, solvent: CDCl3; 13C satellite signals marked by #). The signals at the lower frequency side of H10 presumably arise from
bromine-terminated chains with the TT defect being at the second and third position. The signal marked by an arrow is assigned to the two protons
1
of the TT unit in a TH-TT-HT sequence (TT within the chain). (b) Proton assignments and H NMR end-group analysis of the low molecular
weight defect-free P3HT sample 4a (DP = 16, solvent: CDCl3; 13C satellite signals are marked by #).
the TT unit within the chain, the mechanism of KCTP was
simulated according to a recently presented model.44 Here,
random hopping of the nickel catalyst on the conjugated
backbone together with a stickiness δ when Ni(0) approaches
the chain end is assumed (Figure 2a). The parameter δ
introduces a unidirectional component to the random walk.
The propability to jump to a bromine end group is therefore
biased compared to the otherwise random walk (jump
probabilities to the right and left are 0.5 if Ni(0) is located
somewhere on the chain). The biased probabilities for Ni(0) to
insert at chain ends lead to a higher probability to find Ni there,
i.e., Ni appears to “stick” at the chain ends to a degree
determined by δ. Using this model, the fraction of terminal TT
defects can be simulated with δ being the only parameter. Best
agreement of the model with the experimental values (circles in
Figure 2b) was achieved for δ = 0.55 (line in Figure 2b). In
order to investigate possible influences of the ligand on the
“ring walking” behavior and thus on the distribution of the TT
defect, additional samples were polymerized with Ni(dppe)Cl2.
However, although dppe imparts different reaction rates and
catalytic resting states compared to dppp,56 the fractions of
chains with terminal TT units were always the same for a given
4793
dx.doi.org/10.1021/ja210871j | J. Am. Chem. Soc. 2012, 134, 4790−4805

Journal of the American Chemical Society
Article
Figure 2. (a) Sketch explaining the “stickiness” parameter δ when Ni(0) is close to the chain ends (marked by *). (b) Probability of a terminal TT
defect for Ni(dppp)Cl2-polymerized P3HT as a function of molecular weight (circles) as determined from 1H NMR. The first data point (trimer) is
not from the experiment. The black line results from a simulation with stickiness δ = 0.55. (c) Probability PTT(N,m) to find the TT defect at a certain
monomer index m for three different degrees of polymerization N. (d) Cumulative probability to find the TT defect within the first monomer units n
from the end: ∑PTT (N,n) = ∑i 2PTT (N,n) with i = 1,...,n. The factor 2 in the sum results from the symmetry of the distributions PTT
.
DP (data not shown). For a given δ, the model also gives the
distributions of the TT defect along the chain. Figure 2c
presents these distributions for δ = 0.55 for the three different
degrees of polymerization, DP =17, 40, and 80. In all cases, the
distributions are symmetric and the probability for chains with a
terminal TT unit is highest, but most of the chains have an
internal one. This is clearly illustrated by the cumulative
probability ∑PTT in Figure 2d which gives the probability to
find the TT defect within the first n monomers from one of the
chain ends.
Scheme 2. Structures of Initiators 3a and 3b with Atom
Numbering
3.2. Synthesis of Defect-Free Poly(3-hexylthiophene).
Fully regioregular P3HT chains without any defects are
synthesized via externally initiated KCTP. For this purpose,
new soluble nickel initiators 3 with chelating phosphine ligands
are prepared. 3a,b are synthesized from 2-bromo-3-hexylth-
iophene using diethylbipyridylnickel Et2Ni(bipy) following
ligand exchange57,58 and isolated as golden-orange crystals
(Scheme 1b, Scheme 2).
An alternative route toward initiators 3 starting from 2-
chloromagnesio-3-hexylthiophene and Ni(dppp)Cl2 is feasible
and convenient for the in situ initiation of KCTP;59 however,
we found that workup and purification was easier to handle
using NiEt2bipy. In contrast to conventional KCTP, in which
two transmetalation steps with 2 equiv of active monomer
molecules are required to generate Ni(0), complexes 3a,b
readily release Ni(0) after the first transmetalation and
reductive elimination step. Defect-free P3HT 4a−l was
obtained by adding the active monomer to solutions of the
initiator in THF, polymerizing for indicated times and
quenching the mixture with 5 M HCl.50 3a,b exhibit excellent
solubility in THF, which facilitates precise adjustment of
initiator to monomer ratio. Several batches with different
initiator to monomer ratios were prepared to produce different
molecular weights, and Soxhlet extraction was performed with
methanol only. Thus, narrow polydispersities (GPC) as low as
1.11 were obtained using initiator 3b (see Table 1, for GPC
curves see Supporting Information). The polydispersity indices
obtained from MALDI are shown as well; however, these are
significantly smaller compared to the values from GPC.60 This
4794
dx.doi.org/10.1021/ja210871j | J. Am. Chem. Soc. 2012, 134, 4790−4805

Journal of the American Chemical Society
Article
Figure 3. Molecular weights of the four samples studied in detail. (a) GPC curves measured in THF. (b) MALDI-ToF spectra. GPC curves are
overlaid (dashed lines) by using the scaling MMALDI = 0.62MGPC. In addition to the four samples in (a), the lower molecular weight sample 4c
(0TT-P3HT-26, see Table SI-1) is shown.
Figure 4. (a) Scheme of the semicrystalline microstructure of regioregular P3HT: a, b, and c represent the crystal lattice constants; dc and da are
thicknesses of lamellar crystals and amorphous layers, respectively; Lp = dc + da is the long period. (b) Lorentz-corrected small-angle X-ray scattering
curves at 190 °C for the samples 1TT-P3HT-43 (red), 0TT-P3HT-45 (black), 0TT-P3HT-65 (green), and 1TT-P3HT-76 (blue). All samples were
cooled from the melt in steps of 10 K. Lp as obtained from the peak maximum are shown. (c) Contour length distributions obtained from Figure 3.
Bars indicate Lp values as obtained from (b). (d) Wide-angle scattering curves of the four samples at 190 °C.
common phenomenon is caused by the fact that higher
molecular weights are lower in intensity in MALDI-ToF, which
results in underestimated weight-average molecular weights and
thus smaller PDI values. We will therefore use PDI values
4795
dx.doi.org/10.1021/ja210871j | J. Am. Chem. Soc. 2012, 134, 4790−4805

Journal of the American Chemical Society
Article
Figure 5. (a) Heat flow during cooling from the melt at a rate of 3 K/min. (b) Crystallization onset temperatures for a broader range of molecular
weights and different cooling rates of 3 K/min, 10 K/min, and 20 K/min. Lines are guides to the eye.
obtained from GPC and degrees of polymerization from NMR
throughout this manuscript. Quantitative initiation efficiency of
the initiator is verified by H NMR spectroscopy (Figure 1b).
molecular weights, while the two molecular weights for each
couple are similar. Note that polydispersities are almost equal
for the two couples of comparable molecular weight. Table 1
summarizes all important parameters (for complete character-
ization, see Table SI-1,2). Figure 3a shows GPC curves, and
Figure 3b depicts MALDI ToF spectra. Temperature-depend-
ent combined small-angle and wide-angle X-ray scattering
experiments were performed on bulk samples to characterize
the microstructure. Figure 4a schematically shows the semi-
crystalline morphology of P3HT.46 Alternating crystalline and
amorphous layers form a superstructure with periodicity LP =
dc + da, where LP is the long period. Within the lamellar
crystals, another layered structure is formed, comprising alkyl
chain and thiophene main-chain “sheets” (a-axis). The π−π-
stacking of the planarized backbones is along the b-axis. Figure
4b shows Lorentz-corrected small-angle X-ray (SAXS)
intensities recorded at T = 190 °C, i.e., a temperature close
to the isothermal crystallization temperatures used in the DSC
experiments below. The long periods LP determined from the
peak positions are indicated. The contour-length distributions
shown in Figure 4c together with the LP values have been
calculated from GPC curves for which the molecular weight
axis has been corrected by the MALDI spectra, i.e., the
molecular weight axis of GPC data has been multiplied by a
factor in order to have their maxima at the same positions as
the MALDI spectra (Figure 3b). While samples 1TT-P3HT-43
and 0TT-P3HT-45 have a long period LP very similar to its
contour length lc, the higher molecular weight samples 0TT-
P3HT-65 and 1TT-P3HT-76 exhibit LP values much smaller
than lc. From this we can conclude that 1TT-P3HT-43 and
0TT-P3HT-45 crystallize in an extended-chain conformation,
while for 0TT-P3HT-65 and 1TT-P3HT-76, chain-folding
occurs. Very similar positions of the (100)- and the (020)/
(002)-reflections in Figure 4d indicate that the lattice
parameters are independent of the presence or absence of
the TT defect and are in accordance with the literature
(Table 1).62,63
1
In the spectrum of the fully regioregular chain 4a, the two
doublets of H11 and H12 are clearly seen at 7.16 and 6.93 ppm,
respectively, in accordance with the literature.39,54,55,61 The
fully regioregular structure is further corroborated by the fact
that the signal of the HT-TT triad within the chain, as observed
at 7.00 ppm for conventionally initiated 2a (see arrow in Figure
1a), is absent. While all defect-free polymers carry the 3-
hexylthiophene starting group, the situation concerning the
nature of end-groups is slightly different. In principle, polymers
with H/H termination should be obtained, and this is indeed
observed. With increasing molecular weight, however, the
content of bromine-terminated P3HT chains increases slightly
(see Figure SI-6, Supporting Information). We assume that
bromine end groups arise from an irreversible loss of
coordination of Ni(0) to the conjugated chain, the extent of
which increases with increasing DP. The increasing amount of
bromine terminated chains in 4a−f is consistent with a slight
but increasing low molecular weight tailing observed in the
GPC curves (see Figure SI-5, Supporting Information). The
noticeable intensity at the low molecular weight side, especially
for the higher molecular weight polymers 4e,f thus causes
polydispersity indices (PDI) slightly higher than 1.2. This is not
the case for polymers 4g−l initiated by the dppp-based initiator
3b. These polymers have comparable molecular weights but
exhibit excellent PDIs down to 1.10 for the molecular weight
region investigated. A full table with all relevant molecular
parameters and physical properties is found in the Supporting
Information (Table SI-1,2).
3.3. Influence of Polydispersity and Distributed Tail-
to-Tail Unit on Structure Formation. We select four
polymers for an in-depth analysis in order to reveal the impact
of the TT unit and polydispersity on structure formation and
crystallinity. Samples 2c and 2e initiated with Ni(dppp)Cl2,
denoted as 1TT-P3HT-43 and 1TT-P3HT-76, respectively,
and the two defect-free samples 4i and 4f, denoted as 0TT-
P3HT-45 and 0TT-P3HT-65, respectively, are analyzed by
X-ray scattering and various calorimetric experiments. “1TT”
and “0TT” denote one and no TT defect, respectively, and
the number at the end indicates the degree of polymerization.
The samples synthesized via the same catalyst have different
Figure 5a shows the DSC cooling curves of all four samples
acquired at a rate of 3 K/min. Although the degree of
polymerization of samples 0TT-P3HT-45 and 1TT-P3HT-43 is
almost equal, cf., Table 1, the crystallization onset temperatures
are significantly different. These values are shown in Figure 5b
as crossed symbols. Additionally, the onset temperatures
4796
dx.doi.org/10.1021/ja210871j | J. Am. Chem. Soc. 2012, 134, 4790−4805

Journal of the American Chemical Society
Article
Figure 6. DSC heating curves at a rate of 20 K/min. For each sample, the heating curve after cooling at 3 K/min (black line) is shown in addition to
the heating curves after isothermal crystallization at the two highest temperatures (blue, red lines). (a) 1TT-P3HT-43, (b) 0TT-P3HT-45, (c) 1TT-
P3HT-76, and (d) 0TT-P3HT-65.
obtained from experiments performed with higher cooling rates
and several molecular weights are shown. The onset temper-
atures depended on the cooling rate as expected, i.e., the higher
the cooling rate, the lower the onset temperature. Furthermore,
a clear difference between defect-free samples and samples
containing one TT defect was observed at the same cooling
rates. The P3HT samples with one TT defect crystallized at
lower temperatures as compared to defect-free samples of
similar degree of polymerization. This difference was more
pronounced for lower molecular weights and seemed to vanish
for the highest molecular weights studied. Calorimetric
isothermal crystallization experiments at different temperatures
were used to further explore the observed differences among
the samples. Crystallization at higher temperatures requires
thermodynamically more stable crystals, i.e., polymeric or
oligomeric crystals with a larger thickness dc (cf., Figure 4a).
Thus, size-limiting effects such as polydispersity or potentially
noncrystallizable TT defects might become visible. Figure 6
shows the heating curves after isothermal crystallization at the
two highest temperatures for each sample, in addition to the
curve after nonisothermal crystallization during cooling at 3 K/
min. For all samples, two peaks were observed during melting
after cooling (black line). This is a well-known phenomenon,
namely melting and immediate recrystallization or reorganiza-
tion64 into a thermodynamically more stable form.65 For the
lower molecular weight samples 1TT-P3HT-43 and 0TT-
P3HT-45, heating after isothermal crystallization produced a
single melting peak close to the higher melting peak of the
nonisothermally crystallized samples. This peak shifted to
higher temperatures with increasing crystallization temperature
Tc. Furthermore, a reduction in the enthalpy of melting ΔHm
with increasing Tc was clearly seen. Also for the higher
molecular weight samples 1TT-P3HT-76 and 0TT-P3HT-65,
the dominant melting peak after isothermal crystallization
shifted to higher temperatures with increasing Tc, but here it
was closer to the lower temperature melting peak of the
nonisothermally crystallized case. Furthermore, also for the
isothermally crystallized samples, the second melting peak at
higher temperatures was still observable although significantly
reduced in intensity. Because the results from small-angle X-ray
scattering (Figure 3b and 3c) suggest partial chain folding for
0TT-P3HT-65 and 1TT-P3HT-76, we interpret the higher
temperature melting peaks for these two samples as the melting
of chains that recrystallized in an extended chain configuration.
The melting peak temperatures of all samples from Figure 6
are shown in Figure 7a as a function of Tc. For all four samples,
the increasing melting temperature with increasing Tc indicated
an increased crystal thickness. Variations in melting points
4797
dx.doi.org/10.1021/ja210871j | J. Am. Chem. Soc. 2012, 134, 4790−4805

Journal of the American Chemical Society
Article
Figure 7. (a) Melting peak temperatures as a function of isothermal crystallization temperature (dashed lines guide the eyes). Additionally shown is
the result of a nonisothermal experiment of the defect-free sample 0TT-P3HT-26 (4c, filled triangle). (b) Melting enthalpy after nonisothermal and
isothermal crystallization (filled symbols): For nonisothermal crystallization (−40 K/min: symbols with vertical line; −20 K/min: open symbols,
−10 K/min: half filled symbols; −3 K/min: symbols with cross) enthalpies are plotted as a function of the respective onset temperatures during
cooling. The conversion of the melting enthalpy into crystallinity was performed by assuming 37 J/g for a 100% crystalline sample. (c and d) Same
data as in part b plotted separately for the two couples of lower and higher molecular weight. Lines guide the eyes.
among different samples are more difficult to interpret in terms
of crystal thickness. For instance, Buckley et al. have shown for
low molecular weight poly(ethylene oxide)s that once folded
chain crystals melt at a temperature higher than that of
extended chain crystals of chains having half the molecular
weight.66 One reason for this effect is an entropic contribution
from the chain ends. Such an entropic contribution was
introduced by Flory and Vrij to describe the changes in melting
points of monodisperse n-alkanes.67 It is therefore not evident
that the different melting points of the samples in Figure 7a
directly reflect variations in crystal thickness. Figure 7a
additionally shows the melting point of the defect-free sample
0TT-P3HT-26 (4c) after cooling at 3 K/min (filled triangle).
The onset temperature during cooling was used as the
crystallization temperature. This nonisothermal measurement
together with the isothermal experiments for the other samples
gives an impression on how the melting points change with
molecular weight. It is apparent that it is not straightforward to
relate the observed differences between 0TT-P3HT-45 and
1TT-P3HT-43 in Figure 7a to an effect of the TT defect.
Figure 7b shows ΔHm obtained from heating curves both after
isothermal and nonisothermal crystallization. Figure 7c and 7d
show the same data for the lower and higher molecular weights
separately for clarity. The values of ΔHm were converted into
degrees of crystallinity ϕc = ΔHm/ΔHm∞ by using the recently
suggested value of ΔHm∞ = 37 J/g for the melting enthalpy of a
100% crystalline sample.68 The highest crystallinity of ∼70% is
obtained for the defect-free sample 0TT-P3HT-45. This value
is slightly higher compared to the counterpart with similar
molecular weight 1TT-P3HT-43 having one TT defect, which
exhibited ∼65%. Higher values of ΔHm have only been
reported recently on high pressure crystallized P3HT.37 It is
important to note that irrespective of the presence or absence
of a TT defect in the chain, a decrease in crystallinity is
observed for all samples with increasing crystallization temper-
atures. For 0TT-P3HT-45, a sample without any potentially
noncrystallizable regiodefects that forms extended chain
crystals, polydispersity causes ΔHm to decrease for increasing
Tc: with increasing temperature, crystals require longer chains
in order to be stable, and thus an increasing amount of shorter
chains is not able to crystallize. But for the other three samples,
other reasons such as the presence of the TT defect or a
4798
dx.doi.org/10.1021/ja210871j | J. Am. Chem. Soc. 2012, 134, 4790−4805

Journal of the American Chemical Society
Article
reduction in the fraction of chains that fold might also apply. In
order to be able to separate the contributions from
polydispersity and the TT defect to crystallinity, a simple
model is developed that allows estimation of the crystallinity.
3.4. Development of a Simple Model for the
Determination of Crystallinity. The defect-free P3HT
materials without any potentially noncrystallizable sequence
allow development of a simple model. The model considers a
solid solution of chains of different length, as justified by the
observation of a single long period in Figure 4b. We further
assume that the two repeat units at the chain end are
noncrystalline and that no additional noncrystallizable
sequences within the chains are present. For a given value of
the crystal stem length Nc, three different cases are possible for
a chain with a degree of polymerization N (Figure 8). First, the
molecular weight.34,46 Furthermore, the value chosen for Nf
also affects the calculated crystallinities. P3HT-folds in
monolayers on graphite have been directly observed by
scanning tunneling microscopy,69,70 whereby values of Nf =
6−8 were suggested, which are in accordance with theoretical
models predicting short reentry.69,71 The observed lattice
constant of a = 1.3 nm in these monolayers is smaller than the
value of a = 1.6 nm, which is commonly observed in bulk
samples. This suggests that Nf in the bulk is somewhat larger.
We therefore chose Nf = 8 for all calculations presented here
(for the influence of Nf on the calculated crystallinities, see
Figure SI-7a, Supporting Information). The presented model
neglects all kinetic effects on structure formation but maximizes
the crystallinity per chain under the assumption of a sharply
defined crystal thickness.
The lines in Figure 9b depict the crystallinites ϕc(Nc) as
obtained from the model for the two defect-free samples 0TT-
P3HT-45 and 0TT-P3HT-65. Both calculated crystallinities
exhibit two maxima as a function of Nc. The origin of these two
maxima can be understood as follows. For small values of Nc,
many chains are able to form a fold. With increasing Nc, the
crystalline fraction of these folded chain increases. With further
increasing Nc, the number of chains that are long enough to be
incorporated into the crystal twice decreases. This leads to a
first decrease in crystallinity (labeled “1” in Figure 9b). The
second maximum again arises from an increase in the crystalline
fraction per chain, which is now built into the crystal only once.
Afterward the crystallinity decreases again (“2” in Figure 9b),
because more and more chains are too short to crystallize at all,
as their DP is smaller than the crystal stem length, i.e., N < Nc +
2. In both plots in Figure 9b, the maximum degree of
crystallinity that is achievable is smaller in the first peak (second
peak in Figure 9c and 9d), as chain-folding requires monomer
units which are lost for crystallinity.
Figure 8. Scheme explaining the model for the determination of
crystallinity: A solid solution of crystallizable chains of different length
is assumed, i.e., crystallizable chains of different length are
incorporated into the crystal layers at random positions. Chains
ends (blue filled circles) were assumed to be noncrystalline. The
minimum number of monomers necessary to form a fold is given by
Nf. Nc denotes the crystal stem length, i.e., the thickness of the crystal
layer, in units of monomers.
Figure 9c and 9d show the calculated crystallinities from
Figure 9b as a function of the inverse crystal stem length Nc−1
(gray lines). In addition, the experimentally determined values
ϕc(Tc) are shown. For polymers with a high degree of
polymerization, it is generally observed that the melting
temperature of lamellar crystallites increases with increasing
thickness. This effect is well described by the Gibbs−Thomson
equation Tm = T0 − CNc−1. Here, T0 is the equilibrium melting
point, and the constant C ∼ 2σ/ΔHm∞ is proportional to the
surface energy σ and the melting enthalpy ΔHm∞ of a 100%
crystalline sample.65 For the crystallization temperature and the
crystal thickness, the same relation is often found.65 Although
the above-mentioned entropic contributions from chain ends
and differences in surface energies for extended and folded
chain crystals modify the simple Gibbs−Thompson equation66
for oligomers, experimental data relating temperature and
crystal thickness is usually presented in T−Nc−1diagrams. As
the higher melting points for the increased Tc values in Figure
7a indicate larger crystal thicknesses, we plot the respective x-
axis in Figure 9c and 9d accordingly. The temperature range is
chosen such that the decreasing DSC crystallinities can be
superimposed with the decrease in the calculated crystallinities.
Because 0TT-P3HT-45 exhibits extended-chain crystals, the
experimentally determined crystallinities are superimposed with
the second decrease of the calculated ones (“2” in Figure 9b).
For 0TT-P3HT-65, the comparison of the long period and the
chain length indicates chain-folded crystals, and therefore the
experimentally determined crystallinities are superimposed with
the first decrease of the calculated curve (“1” in Figure 9b).
chain is long enough to fold and can therefore be incorporated
into the crystal twice. In this case, N ≥ 2Nc + 2 + Nf has to
hold, with Nf being the minimum number of monomers
required per fold. Second, if the chain is not long enough to be
incorporated twice, but can be incorporated once, N ≥ Nc + 2
holds. Third, the chain is shorter than the crystal thickness, i.e.,
N < Nc + 2. In this case, the chain is not incorporated into the
crystal.
The crystallinity per chain for these three cases is then given
by 2Nc/N, Nc/N, and 0, respectively. To estimate the
crystallinity of a particular sample for a given value of Nc, the
molecular weight distribution has to be taken into account to
weight the three different cases correctly. The probability p(N)
to find a chain with degree of polymerization N can be directly
estimated from MALDI-ToF. Figure 9a shows the probabilities
p(N) for the three samples 0TT-P3HT-45, 1TT-P3HT-43, and
0TT-P3HT-65. The crystallinity φc(Nc) was estimated
according to:
⎧
N
N
2N
N
c
,
N < 2N + 2 + N
⎪
⎪
c
f
⎨
ϕ (N ) =
p(N)
∑
c
c
⎪
c
N≥N +2
, N ≥ 2N + 2 + N
c
⎪
⎩
c
f
(1)
The model implicitly assumes a rejection of the chains ends
from the crystal. This assumption may be justified by the
observations that the long period of P3HT scales with
4799
dx.doi.org/10.1021/ja210871j | J. Am. Chem. Soc. 2012, 134, 4790−4805

Journal of the American Chemical Society
Article
Figure 9. (a) Distributions of degree of polymerization N from MALDI spectra for the different samples as indicated. (b) Crystallinities calculated
with eq 1 for the two samples 0TT-P3HT-45 and 0TT-P3HT-65. The arrows labeled “1” and “2” indicated the first and second decrease in both
crystallinity curves, respectively, with increasing crystal thickness. (c) Calculated crystallinity (gray line) with p(N) of 0TT-P3HT-45 as a function of
the inverse crystal thickness. The squares are crystallinity values obtained from the melting enthalpies. As indicated by SAXS, samples in this
molecular weight range form extended chain crystals. The temperature axis is therefore plotted such that the decrease in DSC crystallinities
superimposes with the second decrease of the calculated crystallinity (“2” in b). Note that the crystallinity axis has not been adjusted. (d) Same plot
as c for 0TT-P3HT-65. As SAXS indicates once folded chains, the temperature axis is plotted such that the decrease in DSC crystallinities
superimposes with the first decrease in the calculated crystallinity (“1” in b).
Both cases yield very good agreement of calculated and
experimentally determined values. The model also reproduces
the observed slightly lower crystallinity of 0TT-P3HT-65
compared to 0TT-P3HT-45. In addition, the shape of the
experimental data and the calculated curves match well. All this
demonstrates that the very simple model presented here
accounts for the major experimental observations. It also
indirectly supports the recently suggested value ΔHm∞ = 37 J/g
for a 100% crystalline sample,68 as compared to the previously
used value ΔHm∞ = 99 J/g.72 Importantly, the model reveals that
the reasons for the decrease in crystallinity with increasing
isothermal crystallization temperature for 0TT-P3HT-45 and
0TT-P3HT-65 are indeed different: While the decreasing
crystallinity in 0TT-P3HT-45 is caused by the increasing
fraction of chains that are too short to crystallize at all, the
decrease in crystallinity of 0TT-P3HT-65 arises from chains
that are too short to be incorporated into the crystal twice.
The very good agreement of model and experiment
emphasizes the major importance of polydispersity as a limiting
factor for crystallinity. Although the crystallinity of 0TT-P3HT-
45 is rather high, it is essentially limited by polydispersity. Our
model predicts that a reduction in polydispersity should lead to
significantly higher crystallinites. For example, a reduction of
the number of different chain lengths to ∼10 (corresponding to
a PDIMALDI of 1.002) is expected to result in crystallinities as high
as ϕc(Nc) ≈ 0.85 (see Figure SI-7). This is valid for extended-
chain crystals as well as for once-folded ones; however for the
latter, crystallinity is additionally reduced by the fold.
In order to calculate crystallinities of P3HT samples
synthesized via Ni(dppp)Cl2 and to address the question
of whether the TT defect disrupts the crystallizable segment
length thereby limiting crystallinity or not, the simulated
distribution of the TT defect is finally considered in the model.
The equilibrium theory of Flory73 assumes complete exclusion
of defects; however, other theories and experiments suggest
4800
dx.doi.org/10.1021/ja210871j | J. Am. Chem. Soc. 2012, 134, 4790−4805

Journal of the American Chemical Society
Article
Figure 10. (a) Sketch of a P3HT chain with a degree of polymerization of N that contains a TT defect (red circle). (b) 1TT-P3HT-43. Left:
Crystallinity as a function of the crystal thickness calculated considering the TT defect as crystallizable (eq 1, gray line) or as noncrystallizable (eq 2,
dashed orange line). Right: Crystallinity values as obtained from nonisothermal (open circles) and isothermal (filled circles) DSC experiments.
partial inclusion.74,75 Inclusion or exclusion is expected to
depend on the exact nature or the size of the defect,76 because
the crystal lattice is disturbed upon inclusion. In addition,
molecular weight may play a role, too. On the one hand, Ungar
and co-workers found that in crystals composed of ethylene
oligomers with one single methyl branch, the comonomer was
rejected from the crystal.77,78 On the other hand, experiments
on higher molecular weight ethylene-propylene copolymers
have shown that ∼7 propylene units per 100 units can be
incorporated into the crystal, which is accompanied by an
increase in the unit cell dimensions.77,79 The crystal structure of
P3HT with its alternating thiophene and alkyl layers in one
direction, cf., Figure 4a, is rather peculiar, as the alkyl chains are
often amorphous (in the molecular weight range studied
here46) and the ‘crystal’ thus contains disordered regions.
The fact that regiorandom-P3HT, which contains many HH-
and TT-coupling defects, does not crystallize at all, already
demonstrates the detrimental effect that regiodefects can
have on crystallinity. Wu et al. reported that regioregular poly-
(3-butylthiophene-ran-3-octylthiophene), i.e., a thiophene
backbone to which alkyl chains of different length are attached
in a random order, is still able to crystallize.80 This indicates
that a certain irregularity with respect to side-chain order might
be tolerated.
To investigate the behavior of the single TT defect in P3HT,
complete exclusion from the crystal is assumed first. Here only
extended chain crystals are considered. If the TT defect cannot
be incorporated into the crystal, it divides the chain into two
parts, cf., Figure 10a. If for the degree of polymerization N >
2Nc + 2 holds, at least one of those two parts will be longer
than Nc, independent of the location of the TT defect. If N ≤
2Nc + 2, the TT defect has to be located within the first m*-
monomers from either chain end, cf. Figure 10a, for the chain
to be able to crystallize with a crystal stem length of Nc.
Otherwise, both chain segments are shorter than Nc and the
chain will not crystallize at all. The cumulative probability
∑PTT(N,m*) (cf. Figure 2c) gives the fraction of chains with
degree of polymerization N that have the TT defect within the
first m*-monomers from either chain end. Thus, the
crystallinity can be estimated within the model for extended
chain crystals by calculating
N
N
c
ϕ (N ) =
∑
p(N)
c
c
N≥N +2
c
⎧
⎩
1,
N > 2n + 2
c
⎨
∑ P (N; N − N − 1), N ≤ 2n + 2
TT
c
c
(2)
Note that here only extended chain crystals were considered,
and therefore an uncertainty in the choice of the parameter Nf
did not play a role. From this extended model, the crystallinity
expected for 1TT-P3HT-43 is shown in Figure 10b as the
dashed line. For comparison, the crystallinity calculated with eq
1, i.e., assuming complete inclusion of the TT defect, is shown
as a solid line. The crystallinity obtained from the isothermal
DSC experiment at the lowest isothermal crystallization
temperature Tc = 204 °C is ϕc(ΔHm) = 0.59. The calculated
crystallinities for complete exclusion of the TT defects are
lower than this value over the whole range of Nc (see arrow in
Figure 10b).
Thus, for complete exclusion, the predicted reduction in
crystallinity is too large, which in turn indicates that partial
inclusion of the TT defects occurs. Reassessing the thermal
properties discussed in Figure 5b, we can now view the smaller
onset temperatures of crystallization of samples with one TT
unit as an increase in the free energy of the crystal. Complete
inclusion of the TT defect, however, seems unlikely for the
following reason: The decreasing crystallinity of 0TT-P3HT-65
with increasing Tc (Figure 7d) was caused by the increasing
fraction of chains that were too short to be incorporated into
the crystal twice. One would expect that such a decrease occurs
only at higher temperatures for the higher molecular weight
sample 1TT-P3HT-76, if the TT defect could be included
completely. However, as can be seen from Figure 7d, the
crystallinity starts to decrease at lower temperatures for 1TT-
P3HT-76 as compared to 0TT-P3HT-65. This is consistent
with a partial exclusion of the TT unit, and therefore complete
inclusion is unlikely.
Several reports have mentioned that head-to-head (HH)
units do affect crystallinity,28,39,81,82 although a quantitative
4801
dx.doi.org/10.1021/ja210871j | J. Am. Chem. Soc. 2012, 134, 4790−4805

Journal of the American Chemical Society
Article
Figure 11. (a) Temperature-dependent absorption spectra of a 0TT-P3HT-45 film, measured in steps of 10 K. (b) Absorption spectra of 0TT P3HT
with different molecular weights. (c) Comparison of absorption and emission of 0TT-P3HT-45 and 1TT-P3HT-43 (upper graph) and 0TT-P3HT-
65 and 1TT-P3HT-76 (lower graph). (d) Peak ratio of the first (A0‑0) and second (A0‑1) vibronic progression in absorption.
study has not yet been presented. In a very recent report,
Snyder et al. have observed a decrease of crystal thickness with
decreasing RR, which the authors ascribed to noncrystallizable
HH defects that limit the crystallizable segment length.38 Here
it is interesting to note that the crystal lattice of P3HT can
partially tolerate inclusion of the TT defect. Obviously structure
formation is more severely affected by HH defects. Indeed, it is
a reasonable assumption that planarity of the backbone is less
severely distorted by single TT units, whereas HH couplings
cause a stronger torsion of the backbone, either through
sulfur−alkyl chain or alkyl chain−alkyl chain interactions.54
Therefore, the degree of crystallinity of Ni(dppp)Cl2-initiated
samples is only reduced slightly when compared to the defect-
free samples (see Figure 7c).
Finally, it is noteworthy to comment on the utility of the
presented model for other samples in the literature prepared via
Ni(dppp)Cl2. This is a feasible task, provided that high quality
MALDI-ToF spectra are available from polymers that form
extended-chain crystals. If higher molecular weights and/or
broad polydispersities are to be simulated, complication arises
from the fact that multiple chain-folding becomes possible. The
herein presented model is entirely based on thermodynamic
considerations, and we have only treated extended-chain and
once-folded crystals. Structure formation of P3HT with higher
molecular weight and broader PDI can be influenced by kinetic
effects, leading to multiple chain-folding, and a detailed study is
necessary to explore to what extent the presented model can be
applied. There is no intrinsic limitation of the model itself;
however, one must note that for these reasons, no model
predicting crystallinity for high molecular weight polymers is
available at present. To check the validity of the model for
higher molecular weights and larger polydispersities, various
cases that allow for twice or more folded chains need to be
introduced in eq 1 for 0TT-P3HT and in eq 2 for 1TT-P3HT.
3.5. Optical Properties. Our model and the results from
calorimetry both suggest that partial incorporation of the TT
defect takes place in conventionally polymerized P3HT.
Keeping in mind that backbone torsion for TT units is likely
to be smaller in comparison with HH defects,54 and that the
overall degree of crystallinity of 1TT-P3HT-43 is only reduced
slightly compared to 0TT-P3HT-45, the effect of a single TT
unit on the optical properties appears particularly interesting.
Figure 11a shows the absorbance spectra of a 0TT-P3HT-45
film cooled from the melt (T = 290 °C) to room temperature.
The large disorder in the melt causes only one broad
absorption feature at around 450 nm, which is comparable to
its solution spectrum or that of regiorandom amorphous P3HT.
Upon crystallization, the polymer chains undergo conforma-
tional changes and spacial rearrangements that affect the
respective coupling within and among the conjugated polymer
chains.83 On the one hand, the planarization of the polymer
backbone leads to a broad exciton intrachain delocalization, and
on the other hand, the cofacial alignment in the π−π stacking
direction leads to an interchain coupling. The typical resulting
absorption spectrum consists of several distinct vibronic
progressions with a spacing of around 0.18 eV appearing
between 480 and 620 nm. Spano explained these observations
by considering weakly coupled H-aggregates with a strong
exciton−phonon coupling. He calculated the transition
probabilities of the different vibronic bands from the ground
4802
dx.doi.org/10.1021/ja210871j | J. Am. Chem. Soc. 2012, 134, 4790−4805

Journal of the American Chemical Society
Article
state (GS) to an excited state (ES) depending on the disorder
within the aggregates.84 Recrystallization of P3HT-0TT-45
starts from 250 °C and lower (Figure 11a). With higher
crystallinity, the blue shoulder decreases and the ratio between
first and second vibronic level increases, showing increased
ordering. Interestingly, at temperatures below Tc and even
below 60 °C, where no more macroscopic structural changes
take place, further changes are still visible in the optical spectra.
We found these changes to be reversible and there seems to
be a clear temperature dependence for the exciton delocaliza-
tion. These temperature-dependent changes especially have
to be taken into account when performing, for example, time-
resolved pump probe laser spectroscopy, where the pump
laser may heat the sample locally and modulate the spectra.85
The peak ratio of the first and second vibronic transition
A0‑0/A1‑0 can be directly related to the strength of interchain
coupling which again relates to the order and planarization
within H-aggregates.86 We will show later, together with the
photoluminescence of these samples, that this model of pure
H-aggregates may not be simply applicable here. Still, the ratio
A0‑0/A1‑0 directly correlates to the disorder in these polymer
aggregates.84 In the following we plot this ratio which provides
a simple way to compare all samples regarding their inner-
crystalline planarity, i.e., the average conjugation length. Figure 11b
and 11d demonstrates how A0‑0/A1‑0 increases with molecular
weight for defect-free P3HT, which can be explained by the
increasing conjugation length within chain-extended crystals of
increasing thickness. A0−0/A1‑0 approaches unity for the highest
molecular weight sample 0TT-P3HT-65.
of partial inclusion of TT defects and are fully congruent with
the calorimetric data and the calculated crystallinities.
Also photoluminescence (PL) has significant features that
relate to backbone planarization and formation of either J- or
H-aggregates. H-Aggregates have a distinct emission spectra, as
their lowest vibronic transition is forbidden.87 Previously, it has
been shown by temperature-dependent solution measurements
that the first vibronic peak in the emission spectra of P3HT
decreases with increasing aggregation.88 In contrast, Figure 11c
shows that the PL intensity of the E0‑0 peak is very pronounced.
This E0‑0 peak even exceeds the strength of the E1‑0 peak, which
has not been observed previously for P3HT samples with high
regioregularity.88 This observation is even more pronounced
for the defect-free samples. The absorption spectra coincide
with a higher order of the polymer chains, but the strong E0‑0
emission cannot be explained by only assuming H-aggregates.
The increase of the E0‑0 emission would substantiate the picture
of a more delocalized exciton on the polymer chain
experiencing a higher conjugation length, thereby reducing
the coupling between neighboring chains.89 Thus, the defect-
free samples exhibit a clear trend toward longer conjugation
lengths and more planar backbones, which again is in
accordance with partially incorporated TT units in conven-
tionally polymerized chains.
4. CONCLUSIONS
By combining directed synthesis, structure analysis, and
simulations, we have presented an integrated approach to
systematically elucidate the influence of single molecular
parameters on structure formation and optical properties in
semicrystalline poly(3-hexylthiophene). Our results show that
conventionally synthesized P3HT using Ni(dppp)Cl2 as the
catalyst not only contains terminal tail-to-tail defects as believed
to date but also internal ones. The statistical distribution of this
defect within the chain was quantified based on NMR end
group analysis and simulations. Well-defined, narrow-distrib-
uted, defect-free P3HT with a regioregularity of 100% was
prepared via externally initiated KCTP using new, soluble
nickel initiators. The thermal behavior and the optical
properties of the two series of P3HTs, i.e., well-defined
P3HT chains with one distributed tail-to-tail defect obtained
from the conventional initiation via Ni(dppp)Cl2 and well-
defined, entirely defect-free P3HT via new nickel initiators,
were distinctively different. The absence of any potentially
noncrystallizable regiodefect allowed us to propose a simple
model to estimate crystallinities based on the molecular weight
distribution only. Ample viability of this model is demonstrated
by the very good agreement of experimentally determined and
calculated crystallinities. Importantly, polydispersity is identi-
fied as a major parameter determining crystallinity. As-
synthesized defect-free materials exhibit crystallinities as high
as ∼70%, which can further be increased by lowering
polydispersity. The combination of the results on the defect-
free chains, the proposed model, and the distributions of the
regiodefect within Ni(dppp)Cl2-initiated P3HT strongly
indicate that the tail-to-tail unit is at least partially included
into the crystal layers. This result is fully supported by optical
properties, where a high planarity of the backbone is indicated
by the strong increase of the first vibronic absorption peak A0‑0
compared to the second A0‑1. At the same time, the higher
order seems to augment exciton delocalization and to weaken
H-aggregate behavior, which is currently under a more detailed
investigation. Complementary measurements of charge carrier
In principle the TT defect can affect the conjugation length
in two ways: In the case of complete exclusion, a reduction of
the crystallizable sequence length results in thinner crystals and
consequently in a shorter effective conjugation length,
concomitant with a higher amorphous contribution to the
spectra compared to a defect-free sample of identical molecular
weight. In the case of partial inclusion of the TT defect, a
reduction in the planarity of the backbone is likely. This can
limit the conjugation length as well. The comparison of A0‑0/
A1−0 ratios of 1TT-P3HT and 0TT-P3HT materials is shown in
Figure 11c and 11d. In both cases, A0‑0/A1‑0 increases with
molecular weight up to DP ≈ 60, but there is a clear difference
between 0TT and 1TT polymers. The 0TT polymers always
exhibit the higher ratio. The amorphous contribution is,
however, nearly the same for both (Figure 11c). Higher
molecular weights result in thicker crystals and thus in an
increased conjugation length, while at a given molecular weight
the defect-free samples provide more planar and “kink-free”
chains. Both effects increase the exciton delocalization per
chain, i.e., the average effective conjugation length. The higher
ratios A0‑0/A0−1 for defect-free P3HT chains (Figure 11d)
therefore mirror the higher crystallization temperatures (Figure
5b). Further support for the interpretation that the effective
conjugation length within the crystal here is strongly influenced
by the inclusion of TT defects comes from the comparison of
the degrees of polymerization at equal ratios A0‑0/A0‑1. For
instance, in order to achieve a value of A0‑0/A0‑1 = 0.85, the DP
of the 1TT samples have to be nearly twice as high as in the
case of 0TT samples. If only the crystal stem length was
determining the effective conjugation length, this would imply
that the 1TT samples are less crystalline than 0TT samples by a
factor of ∼2 for an identical DP. This is not consistent with the
calorimetric results. Thus, the optical data support the picture
4803
dx.doi.org/10.1021/ja210871j | J. Am. Chem. Soc. 2012, 134, 4790−4805

Journal of the American Chemical Society
Article
(12) Lenes, M.; Wetzelaer, G.-J. A. H.; Kooistra, F. B.; Veenstra,
S. C.; Hummelen, J. C.; Blom, P. W. M. Adv. Mater. 2008, 20, 2116−
2119.
mobility in film and bulk as well as of exciton diffusion lengths
are now needed in order to relate these results to electronic
properties as the next step, and these are currently underway.
Because the presented approach is not limited to P3HT, it is
our hope that it may find application in fundamental studies on
the optoelectronic properties of other conjugated polymers
with semicrystalline morphology, especially where polydisper-
sity can be lowered by employing chain-growth polymerizations
instead of traditional step-growth polycondensations.5,41,90,91
(13) Cheng, Y.-J.; Liao, M.-H.; Chang, C.-Y.; Kao, W.-S.; Wu, C.-E.;
Hsu, C.-S. Chem. Mater. 2011, 23, 4056−4062.
(14) Zhao, G.; He, Y.; Li, Y. Adv. Mater. 2010, 22, 4355−4358.
(15) Bao, Z.; Dodabalapur, A.; Lovinger, A. J. Appl. Phys. Lett. 1996,
69, 4108−4110.
(16) Ihn, K. J.; Moulton, J.; Smith, P. J. Polym. Sci., Part B 1993, 31,
735−742.
(17) Chang, J.-F.; Clark, J.; Zhao, N.; Sirringhaus, H.; Breiby, D. W.;
Andreasen, J. W.; Nielsen, M. M.; Giles, M.; Heeney, M.; McCulloch,
I. Phys. Rev. B 2006, 74, 115318/1−12.
(18) Kline, R. J.; McGehee, M. D.; Kadnikova, E. N.; Liu, J.; Frechet,
J. M. J.; Toney, M. F. Macromolecules 2005, 38, 3312−3319.
(19) Zen, A.; Pflaum, J.; Hirschmann, S.; Zhuang, W.; Jaiser, F.;
Asawapirom, U.; Rabe, J. P.; Scherf, U.; Neher, D. Adv. Funct. Mater.
2004, 14, 757−764.
ASSOCIATED CONTENT
■
S
* Supporting Information
Instrumentation, NMR spectra and additional assignments,
GPC curves, tables with full physical properties, and additional
simulations. This material is available free of charge via the
Internet at http://pubs.acs.org.
(20) Ballantyne, A. M.; Chen, L.; Dane, J.; Hammant, T.; Braun,
F. M.; Heeney, M.; Duffy, W.; McCulloch, I.; Bradley, D. D. C;
Nelson, J. Adv. Funct. Mater. 2008, 18, 2373−2380.
(21) Schilinsky, P.; Asawapirom, U.; Scherf, U.; Biele, M.; Brabec, C.
J. Chem. Mater. 2005, 17, 2175−2180.
AUTHOR INFORMATION
■
Corresponding Author
michael.sommer@makro.uni-freiburg.de
(22) Ma, W.; Kim, J. Y.; Lee, K.; Heeger, A. J. Macromol. Rapid
Commun. 2007, 28, 1776−1780.
Present Address
#Makromolekulare Chemie, Universitat Freiburg, Stefan-Meier-
̈
(23) Koppe, M.; Brabec, C. J.; Heiml, S.; Schausberger, A.; Duffy, W.;
Heeney, M.; McCulloch, I. Macromolecules 2009, 42, 4661−4666.
(24) Kim, J. S.; Lee, Y.; Lee, J. H.; Park, J. H.; Kim, J. K.; Cho, K. Adv.
Mater. 2010, 22, 1355−1360.
Strasse 31, 79100 Freiburg, Germany.
Notes
The authors declare no competing financial interest.
(25) Kim, Y.; Cook, S.; Kirkpatrick, J.; Nelson, J.; Durrant, J. R.;
Bradley, D. D. C.; Giles, M.; Heeney, M.; Hamilton, R.; McCulloch, I.
J. Phys. Chem. C 2007, 111, 8137−8141.
ACKNOWLEDGMENTS
■
The ESRF Grenoble and M. Sztucki are greatly acknowledged
for the provision of synchrotron radiation facilities and
assistance (beamline ID02). This work was carried out with
the support of the Diamond Light Source, UK. V.S., R.T., and
A.K. thank the DFG for financial support (KI-1094/3-1 and KI-
1094/4-1). M.S. is thankful to M. Thelakkat and H. W.
Schmidt, University of Bayreuth, for the opportunity to carry
out preliminary DSC measurements. S.H., R.H.F., M.S., and
W.H. acknowledge the EPSRC for funding (grant number
RG51308). P.K. and U.S. acknowledge the European
Commission (NMP4-SL20010-246123) for funding. S.H. and
P.K. are grateful to J. Clark, University of Cambridge, for fruitful
discussions.
(26) Hiorns, R. C.; de Bettignies, R.; Leroy, J.; Bailly, S.; Firon, M.;
Sentein, C.; Khoukh, A.; Preud’homme, H.; Dagron-Lartigau, C. Adv.
Funct. Mater. 2006, 16, 2263−2273.
̈
(27) Jiang, X. M.; Osterbacka, R.; Korovyanko, O.; An, C. P.;
Horovitz, B.; Janssen, R. A. J.; Vardeny, Z. V. Adv. Funct. Mater. 2002,
12, 587−597.
(28) Kim, Y.; Cook, S.; Tuladhar, S. M.; Choulis, S. A.; Nelson, J.;
Durrant, J. R.; Bradley, D. D. C.; Giles, M.; McCullough, I.; Ha, C. S.;
Ree, M. Nat. Mater. 2006, 5, 197−203.
(29) Woo, C. H.; Thompson, B., C.; Kim, B. J.; Toney, M. F.;
Frechet, J. M. J. J. Am. Chem. Soc. 2008, 130, 16324−16329.
(30) Urien, M.; Bailly, L.; Vignau, L.; Cloutet, E.; de Cuendias, A.;
Wantz, G.; Cramail, H.; Hirsch, L.; Parneix, J. Polym. Int. 2008, 57,
764−769.
(31) Mauer, R.; Kastler, M.; Laquai, F. Adv. Funct. Mater. 2010, 20,
2085−2092.
REFERENCES
■
(32) McMahon, D. P.; Cheung, D. L.; Goris, L.; Dacun, J.; Salleo, A.;
Troisi, A. J. Phys. Chem. B 2011, 115, 19386−19393.
(33) Sirringhaus, H.; Brown, P. J.; Friend, R. H.; Nielsen, M. M.;
Bechgaard, K.; Langeveld-Voss, B. M.W.; Spiering, A. J. H.; Janssen, R.
A. J.; Meijer, E. W.; Herwig, P.; de Leeuw, D. M. Nature 1999, 401,
685−688.
(1) Handbook of Oligo- and Polythiophenes; Fichou, D., Ed.; Wiley-
VCH: Weinheim, 2007.
(2) Heeney, M.; McCulloch, I. Flexible Electronics: Materials and
Applications (Electronic Materials: Science & Technology); Wong, W. S.;
Salleo, A., Eds.; Springer: New York, 2009; pp 261−296.
(3) Katz., H. E.; Bao, Z.; Gilat, S. E. Acc. Chem. Res. 2001, 34, 359−
369.
́
(34) Zhang, R.; Li, B.; Iovu, M. C.; Jeffries-EL, M.; Sauve, G.;
Cooper, J.; Jia, S.; Tristram-Nagle, S.; Smilgies, D. M.; Lambeth, N. D.;
Kowalewski, T; McCullough, R. D. J. Am. Chem. Soc. 2006, 128,
3480−3481.
(4) Salleo, A. Mater. Today 2007, 10, 38−45.
(5) Osaka, I.; McCullough, R. D. Acc. Chem. Res. 2008, 41, 1202−
1214.
(6) Yamamoto, T. NPG Asia Mater. 2010, 2, 54−60.
(7) Dang, M. T.; Hirsch, L.; Wantz, G. Adv. Mater. 2011, 23, 3597−
3602.
(35) Crossland, E.; Rhami, K.; Reiter, G.; Steiner, U.; Ludwigs, S.
Adv. Funct. Mater. 2011, 21, 518−524.
(36) Jimison, L. H.; Toney, M. F.; McCulloch, I.; Heeney, M.; Salleo,
A. Adv. Mater. 2009, 21, 1568−1572.
(8) Padinger, F.; Rittberger, R. S.; Sariciftci, N. S. Adv. Funct. Mater.
2003, 13, 85−88.
(37) Mueller, C.; Zhigadlo, N., D.; Kumar, A.; Baklar, M., A.;
Karpinski, J.; Smith, P.; Kreouzis, T.; Stingelin, N. Macromolecules
2011, 44, 1221−1225.
(9) Ma, W.; Yang, C.; Gong, X.; Lee, K.; Heeger, A. J. Adv. Funct.
Mater. 2005, 15, 1617−1622.
(38) Snyder, C. R.; Henry, J. S.; DeLongchamp, D. M. Macro-
molecules 2011, 44, 7088−7091.
(10) Reyes-Reyes, M.; Kim, K.; Carroll, D. L. Appl. Phys. Lett. 2005,
87, 083506/1−3.
(11) Li, G.; Shrotriya, V.; Huang, J.; Yao, Y.; Moriarty, T.; Emery, K.;
Yang, Y. Nat. Mater. 2005, 4, 864−868.
(39) Mao, H.; Xu, B.; Holdcroft, S. Macromolecules 1993, 26, 1163−
1169.
4804
dx.doi.org/10.1021/ja210871j | J. Am. Chem. Soc. 2012, 134, 4790−4805

Journal of the American Chemical Society
Article
(40) Pankaj, S.; Hempel, E.; Beiner, M. Macromolecules 2009, 42,
716−724.
(77) Ungar, G.; Zeng, X. B.; Brooke, G. M.; Mohammed, S.
Macromolecules 1998, 31, 1875−1879.
(78) Zeng, X.; Xie, F.; Ungar, G. Macromolecules 2007, 40, 5750−
5758.
(41) Yokoyama, A.; Yokozawa, T. Chem. Rev. 2009, 109, 5595−5619.
(42) Iovu, M. C.; Sheina, E. E.; Gil, R. R.; McCullough, R. D.
Macromolecules 2005, 38, 8649−8656.
(79) De Ballesteros, O. R.; Auriemma, F.; Guerra, G.; Corradini, P.
Macromolecules 1996, 29, 7141−7148.
(43) Miyakoshi, R.; Yokoyama, A.; Yokozawa, T. J. Am. Chem. Soc.
2005, 127, 17542−17547.
(44) Tkachov, R.; Senkovskyy, V.; Komber, H.; Sommer, J.-U.; Kiriy,
A. J. Am. Chem. Soc. 2010, 132, 7803−7810.
(45) Zen, A.; Saphiannikova, M.; Neher, D.; Grenzer, J.; Grigorian,
S.; Pietsch, U.; Asawapirom, U.; Janietz, S.; Schreff, U.; Lieberwirth, I.;
Wegner, G. Macromolecules 2006, 39, 2162−2171.
(46) Wu, Z.; Petzold, A.; Henze, T.; Thurn-Albrecht, T.; Lohwasser,
R. H.; Sommer, M.; Thelakkat, M. Macromolecules 2010, 43, 4646−
4653.
(47) Ungar, G.; Zeng, X.-B. Chem. Rev. 2001, 101, 4157−4188.
(48) Zeng, X. B.; Ungar, G. Macromolecules 2001, 34, 6945−6954.
(49) Wilke, G.; Herrmann, G. Angew. Chem., Int. Ed. 1966, 5, 581−
582.
(80) Wu, P.-T.; Ren, G.; Jenekhe, S. A. Macromolecules 2010, 43,
3306−3313.
(81) Souto Maior, R. M.; Hinkelmann, K.; Eckert, H.; Wudl, F.
Macromolecules 1990, 23, 1268−1279.
(82) McCullough, R. D.; Tristram-Nagle, S.; Williams, S. P.; Lowe, R.
D.; Jayaraman, M. J. Am. Chem. Soc. 1993, 115, 4910−4911.
(83) Yang, C.; Orfino, F. P.; Holdcroft, S. Macromolecules 1996, 29,
6510−6517.
(84) Spano, F. C. J. Chem. Phys. 2005, 122, 234701.
(85) Albert-Seifried, S.; Friend, R. H. Appl. Phys. Lett. 2011, 98,
223304.
(86) Clark, J.; Chang, J.; Spano, F. C.; Friend, R. H.; Silva, C. Appl.
Phys. Lett. 2009, 94, 163306.
(87) Spano, F. C. Acc. Chem. Res. 2010, 43, 429−439.
(88) Clark, J.; Silva, C.; Friend, R. H.; Spano, F. Phys. Rev. Lett. 2007,
98, 206406.
(50) Miyakoshi, R.; Yokoyama, A.; Yokozawa, T. Macromol. Rapid
Commun. 2004, 25, 1663−1666.
(51) Lohwasser, R. H.; Thelakkat, M. Macromolecules 2011, 44,
3388−3397.
(89) Theander, M.; Svensson, M.; Ruseckas, A.; Zigmantas, D.;
Sundstrom, V.; Andersson, M. R.; Inganas, O. Chem. Phys. Lett. 2001,
̈
̈
(52) Wu, S.; Huang, L.; Tian, H.; Geng, Y.; Wang, F. Macromolecules
2011, 44, 7558−7567.
337, 227−283.
(90) Kiriy, A.; Senkovskyy, V.; Sommer, M. Macromol. Rapid
Commun. 2011, 32, 1503−1517.
(53) Tkachov, R.; Senkovskyy, V.; Komber, H.; Kiriy, A. Macro-
molecules 2011, 44, 2006−2015.
(91) Senkovskyy, V.; Tkachov, R.; Komber, H.; Sommer, M.;
Heuken, M.; Voit, B.; Huck, W. T. S.; Kataev, V.; Petr, A.; Kiriy, A.
J. Am. Chem. Soc. 2011, 133, 19966−19970.
(54) Barbarella, G.; Bongini, A.; Zambianchi, M. Macromolecules
1994, 27, 3039−3045.
(55) Chen, T.-A.; Wu, X.; Rieke, R. D. J. Am. Chem. Soc. 1995, 117,
233−244.
(56) Lanni, E. L.; McNeil, A. J. Macromolecules 2010, 43, 8039−8044.
(57) Kaul, E.; Senkovskyy, V.; Tkachov, R.; Bocharova, V.; Komber,
H.; Stamm, M.; Kiriy, A. Macromolecules 2010, 43, 77−81.
(58) Senkovskyy, V.; Tkachov, R.; Beryozkina, T.; Komber, H.;
Oertel, U.; Horecha, M.; Bocharova, V.; Stamm, M.; Gevorgyan, S. A.;
Krebs, F. C.; Kiriy, A. J. Am. Chem. Soc. 2009, 131, 16445−16453.
(59) Senkovskyy, V.; Sommer, M.; Komber, H.; Tkachov, R.; Huck,
W.; Kiriy, A. Macromolecules 2010, 43, 10157−10161.
(60) Hanton, S. D. Chem. Rev. 2001, 101, 527−569.
(61) Bras, J.; Pepin-Donat, B. Magn. Reson. Chem. 2001, 39, 57−67.
́
(62) Prosa, T. J.; Winokur, M. J.; Moulton, J.; Smith, P.; Heeger, A. J.
Macromolecules 1992, 25, 4364−72.
(63) Tashiro, K.; Ono, K.; Minagawa, Y.; Kobayashi, M.; Kawai, T.;
Yoshino, K. J. Polym. Sci., Part B: Polym. Phys. 1991, 29, 1223−1233.
(64) Luo, C.-F.; Sommer, J.-U. Phys. Rev. Lett. 2009, 102, 147801.
(65) Strobl, G. The Physics of Polymers, 3rd ed.; Springer: Berlin,
Germany, 2007.
(66) Buckley, C. P.; Kovacs, A. J. Colloid Polym. Sci. 1976, 254, 695−
715.
(67) Flory, P. J.; Vrij, A. J. Am. Chem. Soc. 1963, 85, 3548−3553.
(68) Pascui, O. F.; Lohwasser, R.; Sommer, M.; Thelakkat, M.;
Thurn-Albrecht, T.; Saalwachter, K. Macromolecules 2010, 43, 9401−
̈
9410.
(69) Mena-Osteritz, E.; Meyer, A.; Langeveld-Voss, B. M. W.;
Janssen, R. A. J.; Meijer, E. W.; Bauerle, P. Angew. Chem. Int. 2000, 39,
̈
2680−2684.
́
(70) Grevin, B.; Rannou, P.; Payerne, R.; Pron, A.; Travers, J. P.
J. Chem. Phys. 2003, 118, 7097−7102.
(71) Sommer, J.-U. Eur. Phys. J. E 2006, 19, 413−422.
(72) Malik, S.; Nandi, A. J. J. Polym. Sci., Part B: Polym. Phys. 2002,
40, 2073−2085.
(73) Flory, P. J. Trans. Faraday Soc. 1955, 51, 848−857.
(74) Sanchez, I. C.; Eby, R. K. Macromolecules 1975, 8, 638−641.
(75) Fischer, E. W.; Sterzel, H. J.; Wegner, G. Kolloid-Z. Z. Polym.
1973, 251, 980−990.
(76) Richardson, M. J.; Flory, P. J.; Jackson, J. B. Polymer 1963, 4,
221−236.
4805
dx.doi.org/10.1021/ja210871j | J. Am. Chem. Soc. 2012, 134, 4790−4805